Abstract Functional dissection of neural networks has been revolutionized by the introduction of optogenetic and chemogenetic tools. However, the specificity that can be achieved by currently available tools often falls short, particularly for synapse-specific manipulations. Specifically, although conventional optogenetic and chemogenetic tools are readily applicable to manipulate synaptic transmission by targeting selective subsets of presynaptic inputs or postsynaptic targets, it is often challenging to target synapses based on the unique combinatorial identity of presynaptic inputs and postsynaptic targets. One prominent example is in the nucleus accumbens (NAc), where excitatory glutamatergic inputs from multiple brain regions converge onto principal medium spiny neurons (MSNs). MSNs that express dopamine D1 or D2 receptors (D1 vs. D2 MSNs) receive shared inputs, but drive distinct cellular networks and differentially regulate reward processing. A much desired research tool is to manipulate synaptic transmission at selective subsets of presynaptic inputs onto highly defined postsynaptic neurons (e.g. D1 vs. D2 MSNs). To meet this great challenge, we tested a prototype ?trans-synaptic luminescent opsin? (tsLMO) as a novel approach for highly specific and versatile manipulations of synaptic transmission. TsLMOs consist of postsynaptic luciferase and presynaptic opsins matched to the bioluminescence wavelengths of the luciferase. Upon substrate application, bioluminescence in the post- synaptic neuron acts trans-synaptically to activate the presynaptic opsins, which in turn regulate transmitter release. Thus, the independent expression of the two molecular components of tsLMOs allows for unprece- dented high specificity and versatility for manipulations of synaptic transmission. We have successfully verified the initial efficacy of tsLMOs-based manipulation and will apply it to NAc circuits. We will target the rostral basal lateral amygdala (rBLA)-to-ventral lateral NAc (vlNAc) projection, a subset of the BLA-to-NAc inputs, whose function has not been well understood. Our preliminary results show that this projection regulates natural reward seeking in mice in manners that contrast with current literature. Furthermore, the rBLA-to-vlNAc projection makes monosynaptic excitatory inputs onto both D1 and D2 MSNs, and the two subsets of synapses show differential regulation of glutamate release under physiological conditions. Thus, we propose to use tsLMOs to determine whether rBLA-to-vlNAc transmission onto vlNAc D1 and D2 MSNs differentially regulates natural reward seeking. Expected outcome of this application includes (i) successful development of synaptically targeted tsLMOs for highly specific and versatile manipulations of synaptic transmission; and (ii) an example application of the tsLMO approach for dissecting a specific NAc circuit in reward-elicited behaviors. Given the unique advantages of the tsLMO approach in specificity, versatility, and feasibility, it will become an indispensible tool for functional dissection of neural networks in current neuroscience research. Thus, this application fits the goal of the CEBRA R21 mechanism, and is highly relevant to the mission of the NIH.